Re: A few observations about systemd
On Mon, 18 Jul 2011 at 10:30:15 +0100, Jon Dowland wrote:
> I don't suppose it would be worth maintaining a patch-set in Debian to support
> other OSs: In a hypothetical future where systemd was the default init system
> for Debian, it's probably less work to support multiple init systems and let
> K*BSD/Hurd/*[2] pick another.
I agree with Juliusz' observation that systemd's declarative service
definitions seem sane, and are a reasonable thing to convert into other
inits' native formats (potentially including sysvinit shell scripts) if
required.
I suspect that the shortest path from here to "kFreeBSD can run systemd units"
would be to write one or both of:
* a tool that takes a large subset of systemd unit (service) syntax as input,
and outputs a sysvinit shell script that uses start-stop-daemon (and/or a
new C helper that is run like s-s-d and does some of the same things as
systemd)
* a tool that takes the same command-line parameters as a sysvinit script
and implements them by parsing and running a systemd unit (which would
result in sysvinit scripts that consist of LSB headers, plus one line similar
to "exec not-really-systemd apache2.service "$@")
(In fact, I wonder whether converting daemons' sysvinit scripts into
a declarative format, then running them through a similar tool, would in fact
give us more reliable sysvinit shell scripts than we currently have, even
without replacing sysvinit :-)
S
Reply to: